Rainnny/TextPurify
1
0
Fork 0
Code Issues Pull Requests 1 Actions Activity

Fix punctuation being removed in the replacement

Browse Source
This commit is contained in:
Braydon 2024-06-06 22:28:59 -04:00
parent cc80a10088
commit f5b07e9c12
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
API/src/main/java/me/braydon/profanity/processor/impl/VulgarityProcessor.java
View File

@ -19,6 +19,8 @@ import java.util.regex.Pattern;
* @author Braydon * @author Braydon
*/ */
public final class VulgarityProcessor extends TextProcessor { public final class VulgarityProcessor extends TextProcessor {
private static final String PUNCTUATION_PATTERN = "[\\p{Punct}]*";
/** /**
* Patterns for profane words. * Patterns for profane words.
*/ */
@ -63,8 +65,6 @@ public final class VulgarityProcessor extends TextProcessor {
if (wordPatterns.isEmpty()) { if (wordPatterns.isEmpty()) {
populatePatterns(profanityList); populatePatterns(profanityList);
} }
content = content.replaceAll("\\p{Punct}", ""); // Replace punctuation
// Process single words in the content // Process single words in the content
int offset = 0; int offset = 0;
for (Map.Entry<String, Pattern> entry : wordPatterns.entrySet()) { for (Map.Entry<String, Pattern> entry : wordPatterns.entrySet()) {
@ -78,7 +78,7 @@ public final class VulgarityProcessor extends TextProcessor {
int end = offset + matcher.end(); int end = offset + matcher.end();
String matchedWord = matcher.group(); String matchedWord = matcher.group();
replacement.replace(start, end, Character.toString(replaceChar).repeat(matchedWord.length())); replacement.replace(start, end, Character.toString(replaceChar).repeat(matchedWord.length()));
offset += matchedWord.length() - (end - start); offset+= matchedWord.length() - (end - start);
} }
} }
@ -124,9 +124,9 @@ public final class VulgarityProcessor extends TextProcessor {
for (Character substitution : charSubstitutions.get(lowerChar)) { for (Character substitution : charSubstitutions.get(lowerChar)) {
chars.append(substitution); chars.append(substitution);
} }
obfuscatedWordRegex.append('[').append(chars).append("]+"); obfuscatedWordRegex.append(PUNCTUATION_PATTERN + "[").append(chars).append("]+" + PUNCTUATION_PATTERN);
} else { } else {
obfuscatedWordRegex.append(lowerChar); obfuscatedWordRegex.append(PUNCTUATION_PATTERN).append(lowerChar).append(PUNCTUATION_PATTERN);
} }
} }